From 6dc27b7ca298171d48d3b4d90411b4fbad620770 Mon Sep 17 00:00:00 2001
From: Son Pham-Ba <son.phamba@epfl.ch>
Date: Mon, 6 Nov 2023 17:20:36 +0100
Subject: [PATCH] add tests for io/find ges_disc

---
 gpm_api/io/ges_disc.py             | 19 +++++-----
 gpm_api/tests/test_io/test_find.py | 56 +++++++++++++++++++++++++++---
 2 files changed, 62 insertions(+), 13 deletions(-)

diff --git a/gpm_api/io/ges_disc.py b/gpm_api/io/ges_disc.py
index e17f8d9a..bfc159ed 100644
--- a/gpm_api/io/ges_disc.py
+++ b/gpm_api/io/ges_disc.py
@@ -5,7 +5,6 @@
 @author: ghiggi
 """
 import datetime
-import os
 import re
 import subprocess
 
@@ -51,7 +50,7 @@ def _get_gesc_disc_list_path(url):
     list_content = [s for s in list_content if s != ""]
     if len(list_content) == 0:
         raise ValueError(f"The GES DISC {url} directory is empty.")
-    list_path = [os.path.join(url, s) for s in list_content]
+    list_path = [f"{url}/{s}" for s in list_content]
     return list_path
 
 
@@ -72,11 +71,11 @@ def _get_gesc_disc_list_path(url):
 def _get_ges_disc_server(product):
     # TRMM
     if is_trmm_product(product):
-        ges_disc_base_url = "https://disc2.gesdisc.eosdis.nasa.gov/data/"
+        ges_disc_base_url = "https://disc2.gesdisc.eosdis.nasa.gov/data"
 
     # GPM
     else:
-        ges_disc_base_url = "https://gpm1.gesdisc.eosdis.nasa.gov/data"
+        # ges_disc_base_url = "https://gpm1.gesdisc.eosdis.nasa.gov/data"
         ges_disc_base_url = "https://gpm2.gesdisc.eosdis.nasa.gov/data"
     return ges_disc_base_url
 
@@ -114,9 +113,11 @@ def _get_ges_disc_product_directory_tree(product, date, version):
 
     # Specify the directory tree
     # --> TODO: currently specified only for L1 and L2
-    directory_tree = os.path.join(
-        folder_name,
-        datetime.datetime.strftime(date, "%Y/%j"),
+    directory_tree = "/".join(
+        [
+            folder_name,
+            datetime.datetime.strftime(date, "%Y/%j"),
+        ]
     )
     return directory_tree
 
@@ -148,7 +149,7 @@ def get_ges_disc_product_directory(product, date, version):
         product=product, date=date, version=version
     )
     # Define product directory where data are listed
-    url_product_dir = os.path.join(url_server, dir_structure)
+    url_product_dir = f"{url_server}/{dir_structure}"
     return url_product_dir
 
 
@@ -248,5 +249,5 @@ def define_gesdisc_filepath(product, product_type, date, version, filename):
     # Retrieve product directory url
     url_product_dir = get_ges_disc_product_directory(product=product, date=date, version=version)
     # Define GES DISC filepath
-    fpath = os.path.join(url_product_dir, filename)
+    fpath = f"{url_product_dir}/{filename}"
     return fpath
diff --git a/gpm_api/tests/test_io/test_find.py b/gpm_api/tests/test_io/test_find.py
index 7bfa6976..6f109259 100644
--- a/gpm_api/tests/test_io/test_find.py
+++ b/gpm_api/tests/test_io/test_find.py
@@ -1,6 +1,6 @@
 from datetime import datetime
 import os
-from typing import Dict
+from typing import Dict, List
 
 from pytest_mock.plugin import MockerFixture
 
@@ -12,7 +12,7 @@ def test_get_local_daily_filepaths(
     mock_configuration: Dict[str, str],
     mocker: MockerFixture,
     product_info: Dict[str, dict],
-):
+) -> None:
     """Test _get_all_daily_filepaths for "local" storage"""
 
     storage = "local"
@@ -82,7 +82,7 @@ def test_get_local_daily_filepaths(
 def test_get_pps_daily_filepaths(
     mocker: MockerFixture,
     product_info: Dict[str, dict],
-):
+) -> None:
     """Test _get_all_daily_filepaths for "pps" storage"""
 
     stoarge = "pps"
@@ -94,7 +94,7 @@ def test_get_pps_daily_filepaths(
         "file2.HDF5",
     ]
 
-    def mock_get_pps_file_list(url_product_dir):
+    def mock_get_pps_file_list(url_product_dir: str) -> List[str]:
         # Remove the base URL, assuming they have the followgin format:
         # RS: https://arthurhouhttps.pps.eosdis.nasa.gov/text/...
         # NRT: https://jsimpsonhttps.pps.eosdis.nasa.gov/text/...
@@ -187,3 +187,51 @@ def mock_get_pps_file_list(url_product_dir):
         )
         expected_filepaths = [f"{base_url}{filename}" for filename in mock_filenames]
         assert returned_filepaths == expected_filepaths
+
+
+def test_get_gesdisc_daily_filepaths(
+    mocker: MockerFixture,
+    product_info: Dict[str, dict],
+) -> None:
+    """Test _get_all_daily_filepaths for "ges_disc" storage"""
+
+    stoarge = "ges_disc"
+    date = datetime(2020, 12, 31)
+    version = 7
+
+    # Mock gpm_api.io.ges_disc._get_gesc_disc_list_path, which uses wget to get a list of files
+    mock_filenames = [
+        "file1.HDF5",
+        "file2.HDF5",
+    ]
+
+    def mock_get_gesc_disc_list_path(url: str) -> List[str]:
+        return [f"{url}/{filename}" for filename in mock_filenames]
+
+    mocker.patch(
+        "gpm_api.io.ges_disc._get_gesc_disc_list_path", side_effect=mock_get_gesc_disc_list_path
+    )
+
+    for product, info in product_info.items():
+        version = info["available_versions"][-1]
+        ges_disc_dir = info["ges_disc_dir"]
+        if ges_disc_dir is None:
+            continue
+
+        returned_filepaths = find._get_all_daily_filepaths(
+            storage=stoarge,
+            date=date,
+            product=product,
+            product_type=None,
+            version=version,
+            verbose=True,
+        )
+
+        if "TRMM" in ges_disc_dir:
+            subdomain = "disc2"
+        else:
+            subdomain = "gpm2"
+
+        base_url = f"https://{subdomain}.gesdisc.eosdis.nasa.gov/data/{ges_disc_dir}.0{version}/{date.strftime('%Y/%j')}"
+        expected_filepaths = [f"{base_url}/{filename}" for filename in mock_filenames]
+        assert returned_filepaths == expected_filepaths